Binary bet trading
Hi
I'm a software developer and have developed an application platform for trading Ftse100 up down bets.
The development of this application has essentially come to an end.
I have decided to contribute some of the knowledge gained to a forum in the hope of assisting in the trading strategies of like minded traders.
There are obviously some areas that I am unable to comment upon namely the specifics of key algorithms. However I will generalise. This is because the algorithms are being further developed for use in other trading strategies/approaches.
My background:
Educated to A level. ( in the days when A-levels meant something)
Open University maths degree (not completed) mainly concentrating on probability & non linear dynamics.
Between the age of 21 – 30 I was working as an accountant
My spare time was spent developing and testing statistical models of stock market behaviour. By the age of 26 I had developed a basic reliable algorithm that seemed to work.
Between 1998 – 2000 I traded the algorithm (a very simplified version) and made enough money to pay off the mortgage.
re-trained as a software developer.
Between 2001 – 2004 I worked as a software developer on military systems.
In early 2003 I tried some manual binary betting (without any software system in place) resulting in highly varied results (that is I won then lost it all back !)
By December 2003 myself and a colleague had developed a basic system which was able to grab real time data from the binary bet web site analyse and generate trading signals. The signals were emailed and sent via msn to us.
The software has gone through various versions to the current and final version v 3.2. the changes have essentially been in revised logic rules and improvements to algorithms. There have been no changes to rules or algorithms for approx 14 mths.
The software has a success rate of 83% . Namely following the signals results in 83% of the trades being closed at a profit.
The software was developed with the idea of freeing myself from the drudgery of having to work for someone else. The objective was to sell up move abroad and trade from a pleasant sunny climate.
Two factors have conspired to block the objective . The first is I actually enjoy my job and earn enough that I don’t want to leave it. The second is despite the success of the system I couldn’t stop myself from second guessing it. This caused an immense amount of stress./ addiction like feeding a fruit machine !
The last continuous period of trading was 13 weeks leading up to the London bombings. The account was opened with £1000 and before the bombings stood at approx £1600.
Average bet was approx £3/pt.
The application is not for sale.
The first area of concern when developing the system was being able to grab the binary bet prices. We were able to do this via a little know back door into the Bin bet web site. This is the link
LINK Removed that stored FTSE up down data. (the link has now been changed ie it doesn’t store ftse data but you can obtain currency and commodity data from it) You could essentially get their prices by loading the web page address into an application and simply reading the html file. The link produces a static page , but keep refreshing it and the data is updated. This is what we did in code, kept calling that page 1020 times a day to be precise J.
In actual fact all this information really gives a trader is the knowledge that when they get a signal they know what the current bin bet price is. However by the time you’ve logged into the web site to place the trade it’s historic which means you could get a marginally better or worse price. Being able to grab a full days data allowed us to back engineer the pricing algorithm. Essentially this is a time decay function with the movement of the index from the open (+/- amount) superimposed along with some function that prices in volatility .
Whether there underlying model is based on the futures price was irrelevant in developing our own model of bin bet prices. What we wanted was a model that gets us pretty close to the Bin bet pricing model. Being able to grab the data from the site enabled us to do this.
We now have about 300 + days of actual bin bet & FTSE 100 Up Down price data, that we can use for testing various strategies against
Obviously they could always change their model. But doing so could have a fairly severe impact on their profitability especially in volatile markets. Plus as long as market behaviour stays the same then I can’t see the pricing model changing much.
Our pricing model isn’t an exact match of Bin bet price model. We are able to to get within 2% of bin bet price for 90% of prices, but we can diverge by as much as 10% of the bin bet price when market volatility increases. We suspect that the bin bet volatility function is protecting them from opportunistic traders by frightening them off with extreme prices, or limiting any profits. However we found that during high periods of volatility especially in the morning or early afternoon our model might be indicating a price of 78 when the bin bet price might be closer to 83 or 84. These differences completely disappear as the market approaches close. And bin bet price and our model merge.
When developing our price we would often get prices above 100. At first we thought our model was wrong, but we think that bin bet get similar pricing. However when their sys generates a price above 100 they just display 100 . So they have a rule something like.
If (nUpbuyPrice > 100)
{
nUpbuyPrice = 100;
Display (nUpBuyPrice);
}
The trade signalling is based on determining high probability turning points using non-linear functions. That’s really as far as I can go on the signalling. The ability to know when a significant turn is coming is key to the systems success (especially in volatile markets). Our difficulty was in determining magnitude ie by how much the index was likely to move. This represented a significant hurdle and left exposed to some volatile reading periods. However we believe we may have found a solution that is being developed.
Hope this helps, encourages enlightens or amuses J
(which also has some extra fields such as time updated and the change)
